U.S. patent number 10,794,979 [Application Number 15/780,658] was granted by the patent office on 2020-10-06 for removal of image artifacts in sense-mri.
This patent grant is currently assigned to Koninklijke Philips N.V.. The grantee listed for this patent is KONINKLIJKE PHILIPS N.V.. Invention is credited to Peter Boernert, Ivan Dimitrov, Miha Fuderer.
United States Patent |
10,794,979 |
Boernert , et al. |
October 6, 2020 |
Removal of image artifacts in sense-MRI
Abstract
The invention provides for a magnetic resonance imaging system
(100, 300) comprising: a radio-frequency system (116, 122, 124,
126, 126', 126'', 126''') for acquiring magnetic resonance data
(152) from an imaging zone (108), wherein the radio-frequency
system comprises multiple antenna elements (126, 126', 126'',
126'''); a memory (140) containing machine executable instructions
(170) and pulse sequence commands (150), wherein the pulse sequence
commands cause the processor to acquire magnetic resonance data
from the multiple antenna elements according to a SENSE protocol;
and a processor. Execution of the machine executable instructions
causes the processor to: control (200) the magnetic resonance
imaging system with the pulse sequence commands to acquire the
magnetic resonance data; reconstruct (202) a preliminary image
(154) using the magnetic resonance imaging data; calculate (204) a
fit (159) between an anatomical model (156) and the preliminary
image, wherein the anatomical model comprises a motion likelihood
map (158); identify (206) at least one image artifact origin (160)
at least partially using the motion likelihood map and the fit;
determine (208) an extended SENSE equation (162) at least partially
using at least one image artifact origin; and construct (210) a
corrected SENSE image (164) using the extended SENSE equation.
Inventors: |
Boernert; Peter (Hamburg,
DE), Fuderer; Miha (Best, NL), Dimitrov;
Ivan (Bothell, WA) |
Applicant: |
Name |
City |
State |
Country |
Type |
KONINKLIJKE PHILIPS N.V. |
Eindhoven |
N/A |
NL |
|
|
Assignee: |
Koninklijke Philips N.V.
(Eindhoven, NL)
|
Family
ID: |
1000005096988 |
Appl.
No.: |
15/780,658 |
Filed: |
November 8, 2016 |
PCT
Filed: |
November 08, 2016 |
PCT No.: |
PCT/EP2016/076908 |
371(c)(1),(2),(4) Date: |
June 01, 2018 |
PCT
Pub. No.: |
WO2017/092973 |
PCT
Pub. Date: |
June 08, 2017 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20180356484 A1 |
Dec 13, 2018 |
|
Related U.S. Patent Documents
|
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
Issue Date |
|
|
62262706 |
Dec 3, 2015 |
|
|
|
|
Foreign Application Priority Data
|
|
|
|
|
Jan 14, 2016 [EP] |
|
|
16151326 |
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G01R
33/56509 (20130101); G01R 33/5611 (20130101) |
Current International
Class: |
G01R
33/561 (20060101); G01R 33/565 (20060101) |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
|
|
|
|
|
|
|
2008010126 |
|
Jan 2008 |
|
WO |
|
2014087270 |
|
Jun 2014 |
|
WO |
|
2014154544 |
|
Oct 2014 |
|
WO |
|
Other References
"Bernstein Handbook of MRI Pulse Sequences" 2004 Section 13.3.
cited by applicant .
Winkelmann et al "Ghost Artifact Removal Using a Parallel Imaging
Approach" Magn. Reson. Med. Oct. 2005; 54(4) p. 1002-9. cited by
applicant .
Alexey A. Samsonov et al: "POCS-enhanced correction of motion
artifacts in parallel MRI", Magnetic Resonance in Medicine,vol. 53,
No. 4, Apr. 1, 2010 (Apr. 1, 2010),pp. 1104-1110. cited by
applicant .
Winkelmann R et al:"Residual aliasing removal in higher order
SENSE",Proceedings of the International Society for Magnetic
Resonance in Medicine, ISMRM, 13th Scientific Meeting and
Exhibition, Miami Beach, Florida, USA, May 7-13, 2005, Apr. 23,
2005 (Apr. 23, 2005), p. 2426. cited by applicant .
Martin Uecker et al: "ESPIRiT--an eigenvalue approach to
autocalibrating parallel MRI: Where SENSE meets GRAPPA", Magnetic
Resonance in Medicine., vol. 71, No. 3, May 6, 2013 (May 6, 2013),
pp. 990-1001. cited by applicant .
Chu "POCs-Based Reconstruction of Multiplexed Sensitivity Encoded
MRI (POSCMUSE): A General Algorithm for Reducing Motion Related
Artifacts" Magn. Reson. in Med. (2014). cited by applicant.
|
Primary Examiner: Le; Son T
Claims
The invention claimed is:
1. A magnetic resonance imaging system comprising: a
radio-frequency system for acquiring magnetic resonance data from
an imaging zone, wherein the radio-frequency system comprises
multiple antenna elements; a memory containing machine executable
instructions and pulse sequence commands, wherein the pulse
sequence commands cause the processor to acquire magnetic resonance
data from the multiple antenna elements according to a SENSE
protocol; a processor, wherein execution of the machine executable
instructions causes the processor to: control the magnetic
resonance imaging system with the pulse sequence commands to
acquire the magnetic resonance data; reconstruct a preliminary
image using the magnetic resonance imaging data; calculate a fit
between an anatomical model and the preliminary image, wherein the
anatomical model comprises a motion likelihood map; identify at
least one image artifact origin at least partially using the motion
likelihood map and the fit; determine an extended SENSE equation at
least partially using at least one image artifact origin; and
construct a corrected SENSE image using the extended SENSE
equation.
2. The magnetic resonance imaging system of claim 1, wherein
execution of the machine executable instructions causes the
processor to: reconstruct a measured coil image for each of the
multiple antenna elements using the magnetic resonance data;
construct by using a set of coil sensitivities to combine the
measured coil image of each of the multiple antenna elements
according to the SENSE protocol.
3. The magnetic resonance imaging of claim 2, wherein the
preliminary image comprises the preliminary SENSE image.
4. The magnetic resonance imaging system of claim 2 wherein
execution of the machine executable instructions further causes the
processor to: construct a backprojected image for each of the
multiple antenna elements using the preliminary SENSE image and the
coil sensitivities; and compare the backprojected image to the
measured coil image for each of the multiple antenna elements to
identify a set of affected voxels for each of the multiple antenna
elements; wherein the identification of the at least one image
artifact origin is performed in image space; wherein the
identification of the at least one image artifact origin is
performed at least partially using the set of affected voxels at
least partially using the motion likelihood map and the fit.
5. The magnetic resonance imaging system of claim 4, wherein the at
least one image artifact origin is corrected by numerically
searching for a maximum of a consistency measure within a
predetermined vicinity of each of the at least one image artifact
origin before constructing the corrected SENSE image, wherein the
consistency measure is dependent upon the difference between the
set of affected voxels in the preliminary SENSE image and
backprojected trial SENSE images for each of the multiple antenna
elements, wherein the backprojected trial sense images are
constructed from a trial SENSE image, wherein the trial sense image
is constructed using a trial SENSE equation.
6. The magnetic resonance imaging system of claim 5, wherein the
trial SENSE equation that maximizes the consistency measure is the
extended SENSE equation.
7. The magnetic resonance imaging system of claim 4, wherein
execution of the machine executable instructions further causes the
processor to modify the at least one image artifact origin by
registering the set of affected voxels to the preliminary
image.
8. The magnetic resonance imaging system of claim 1, wherein the
extended SENSE equation is chosen to minimize a contribution from
at least a portion of the at least one image artifact origin.
9. The magnetic resonance imaging system of claim 1, wherein the
preliminary image comprises a survey scan image.
10. The magnetic resonance imaging system of claim 1, wherein the
at least one image artifact origin is two-dimensional or
three-dimensional.
11. A computer program product comprising machine executable
instructions for execution by a processor controlling a magnetic
resonance imaging system, wherein the magnetic resonance imaging
system comprises a radio-frequency system for acquiring magnetic
resonance data from an imaging zone, wherein the radio-frequency
system comprises multiple antenna elements, wherein execution of
the machine executable instructions causes the processor to:
control the magnetic resonance imaging system with pulse sequence
commands to acquire the magnetic resonance data, wherein the pulse
sequence commands cause the processor to acquire magnetic resonance
data from the multiple antenna elements according to a SENSE
protocol; reconstruct a preliminary image using the magnetic
resonance imaging data; calculate a fit between an anatomical model
and the preliminary image, wherein the anatomical model comprises a
motion likelihood map; identify at least one image artifact origin
at least partially using the motion likelihood map and the fit;
determine an extended SENSE equation at least partially using at
least one image artifact origin; and construct a corrected SENSE
image using the extended SENSE equation.
12. A method of operating a magnetic resonance imaging system,
wherein the magnetic resonance imaging system comprises a
radio-frequency system for acquiring magnetic resonance data from
an imaging zone, wherein the radio-frequency system comprises
multiple antenna elements, wherein the method comprises the steps
of: controlling the magnetic resonance imaging system with pulse
sequence commands to acquire the magnetic resonance data, wherein
the pulse sequence commands cause the processor to acquire magnetic
resonance data from the multiple antenna elements according to a
SENSE protocol; reconstructing a preliminary image using the
magnetic resonance imaging data; calculating a fit between an
anatomical model and the preliminary image, wherein the anatomical
model comprises a motion likelihood map; identifying at least one
image artifact origin at least partially using the motion
likelihood map and the fit; determining an extended SENSE equation
at least partially using at least one image artifact origin; and
constructing a corrected SENSE image using the SENSE equation.
13. The method of claim 12, wherein the method further comprises:
reconstructing a measured coil image for each of the multiple
antenna elements using the magnetic resonance data; constructing a
preliminary SENSE image by using a set of coil sensitivities to
combine the measured coil image for each of the multiple antenna
elements according to the SENSE protocol; constructing a
backprojected image for each of the multiple antenna elements using
the preliminary SENSE image and the coil sensitivities; and
comparing the backprojected image to the measured coil image for
each of the multiple antenna elements to identify a set of affected
voxels for each of the multiple antenna elements; wherein the
identification of the at least one image artifact origin is
performed in image space; wherein the identification of the at
least one image artifact origin is performed at least partially
using the set of affected voxels at least partially using the
motion likelihood map and the fit or registration, wherein the at
least one image artifact origin is corrected by numerically
searching for a maximum of a consistency measure within a
predetermined vicinity of each of the at least one image artifact
origin before constructing the corrected SENSE image, wherein the
consistency measure is dependent upon the difference between the
set of affected voxels in the preliminary SENSE image and
backprojected trial SENSE images for each of the multiple antenna
elements, wherein the backprojected trial sense images are
constructed from a trial SENSE image, wherein the trial sense image
is constructed using a trial SENSE equation.
14. The method of claim 12, wherein the extended SENSE equation
comprises an extended coil sensitivity matrix, wherein the extended
coil sensitivity matrix is chosen to minimize a contribution from
at least a portion of the at least one image artifact origin.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
This application is a U.S. national phase application of
International Application No. PCT/EP2016/076908, filed on Nov. 8,
2016, which claims the benefit of U.S. provisional Application Ser.
No. 62/262,706 filed Dec. 3, 2015 and EP Application Serial No.
16151326.2 filed on Jan. 14, 2016 each of which is incorporated
herein by reference.
TECHNICAL FIELD OF THE INVENTION
The invention relates to magnetic resonance imaging, in particular
to the removal of artifacts during SENSE magnetic resonance imaging
protocols.
BACKGROUND OF THE INVENTION
A large static magnetic field is used by Magnetic Resonance Imaging
(MRI) scanners to align the nuclear spins of atoms as part of the
procedure for producing images within the body of a patient. This
large static magnetic field is referred to as the B0 field.
During an MRI scan, Radio Frequency (RF) pulses generated by a
transmitter antenna or antenna element cause perturbations to the
local magnetic field, and RF signals emitted by the nuclear spins
are detected by a receiver antenna or an array of antenna elements.
These RF signals are used to construct the MRI images. These
antennas or antenna elements can also be referred to as coils. The
term coil is often used interchangeably to describe either an
antenna or an antenna element. Further, the transmitter and
receiver antennas can also be integrated into a single transceiver
antenna that performs both functions. It is understood that the use
of the term transceiver antenna also refers to systems where
separate transmitter and receiver antennas are used. The
transmitted RF field is referred to as the B1 field. During longer
scan the subject can have internal or external motion which
corrupts the data and results in images with blurs or
artifacts.
SENSE is a parallel imaging technique. In parallel imaging
techniques multiple antenna elements are used to acquire data
simultaneously. Coil sensitivity maps (CSM) contains spatial
sensitivity of all the antenna elements. In this case a "coil"
refers to an antenna element. The coil sensitivity maps are used to
combine the data acquired using each of the individual antenna
elements into a single composite image. SENSE greatly accelerates
the acquisition of the magnetic resonance image. Magnetic resonance
parallel-imaging reconstruction techniques are briefly outlined in
section 13.3 of "the handbook of MRI Pulse Sequences" by Bernstein
et al. published by Elsevier Academic Press, 2004 (hereafter
Bernstein et. al.)
The journal article Winkelmann et. al., "Ghost artifact removal
using a parallel imaging approach," Magn. Reson. Med. 2005 October;
54(4):1002-9 (hereafter Winkelmann et. al. describes a ghost
artifact removal algorithm that uses parallel imaging. An extended
SENSE formulation is used to remove ghosting artifacts. An extended
SENSE reconstruction is determined by numerically trying different
origins for the ghosting artifact and ranking them using a
consistency measure. A disadvantage of this method is that it is
very numerically intensive.
SUMMARY OF THE INVENTION
The invention provides for a magnetic resonance imaging system, a
computer program product, and a method of operating the magnetic
resonance imaging system in the independent claims. Embodiments are
given in the dependent claims.
As will be appreciated by one skilled in the art, aspects of the
present invention may be embodied as an apparatus, method or
computer program product. Accordingly, aspects of the present
invention may take the form of an entirely hardware embodiment, an
entirely software embodiment (including firmware, resident
software, micro-code, etc.) or an embodiment combining software and
hardware aspects that may all generally be referred to herein as a
"circuit," "module" or "system." Furthermore, aspects of the
present invention may take the form of a computer program product
embodied in one or more computer readable medium(s) having computer
executable code embodied thereon.
Any combination of one or more computer readable medium(s) may be
utilized. The computer readable medium may be a computer readable
signal medium or a computer readable storage medium. A
`computer-readable storage medium` as used herein encompasses any
tangible storage medium which may store instructions which are
executable by a processor of a computing device. The
computer-readable storage medium may be referred to as a
computer-readable non-transitory storage medium. The
computer-readable storage medium may also be referred to as a
tangible computer readable medium. In some embodiments, a
computer-readable storage medium may also be able to store data
which is able to be accessed by the processor of the computing
device. Examples of computer-readable storage media include, but
are not limited to: a floppy disk, a magnetic hard disk drive, a
solid state hard disk, flash memory, a USB thumb drive, Random
Access Memory (RAM), Read Only Memory (ROM), an optical disk, a
magneto-optical disk, and the register file of the processor.
Examples of optical disks include Compact Disks (CD) and Digital
Versatile Disks (DVD), for example CD-ROM, CD-RW, CD-R, DVD-ROM,
DVD-RW, or DVD-R disks. The term computer readable-storage medium
also refers to various types of recording media capable of being
accessed by the computer device via a network or communication
link. For example a data may be retrieved over a modem, over the
internet, or over a local area network. Computer executable code
embodied on a computer readable medium may be transmitted using any
appropriate medium, including but not limited to wireless, wire
line, optical fiber cable, RF, etc., or any suitable combination of
the foregoing.
A computer readable signal medium may include a propagated data
signal with computer executable code embodied therein, for example,
in baseband or as part of a carrier wave. Such a propagated signal
may take any of a variety of forms, including, but not limited to,
electro-magnetic, optical, or any suitable combination thereof. A
computer readable signal medium may be any computer readable medium
that is not a computer readable storage medium and that can
communicate, propagate, or transport a program for use by or in
connection with an instruction execution system, apparatus, or
device.
`Computer memory` or `memory` is an example of a computer-readable
storage medium. Computer memory is any memory which is directly
accessible to a processor. `Computer storage` or `storage` is a
further example of a computer-readable storage medium. Computer
storage is any non-volatile computer-readable storage medium. In
some embodiments computer storage may also be computer memory or
vice versa.
A `processor` as used herein encompasses an electronic component
which is able to execute a program or machine executable
instruction or computer executable code. References to the
computing device comprising "a processor" should be interpreted as
possibly containing more than one processor or processing core. The
processor may for instance be a multi-core processor. A processor
may also refer to a collection of processors within a single
computer system or distributed amongst multiple computer systems.
The term computing device should also be interpreted to possibly
refer to a collection or network of computing devices each
comprising a processor or processors. The computer executable code
may be executed by multiple processors that may be within the same
computing device or which may even be distributed across multiple
computing devices.
Computer executable code may comprise machine executable
instructions or a program which causes a processor to perform an
aspect of the present invention. Computer executable code for
carrying out operations for aspects of the present invention may be
written in any combination of one or more programming languages,
including an object oriented programming language such as Java,
Smalltalk, C++ or the like and conventional procedural programming
languages, such as the "C" programming language or similar
programming languages and compiled into machine executable
instructions. In some instances the computer executable code may be
in the form of a high level language or in a pre-compiled form and
be used in conjunction with an interpreter which generates the
machine executable instructions on the fly.
The computer executable code may execute entirely on the user's
computer, partly on the user's computer, as a stand-alone software
package, partly on the user's computer and partly on a remote
computer or entirely on the remote computer or server. In the
latter scenario, the remote computer may be connected to the user's
computer through any type of network, including a local area
network (LAN) or a wide area network (WAN), or the connection may
be made to an external computer (for example, through the Internet
using an Internet Service Provider).
Aspects of the present invention are described with reference to
flowchart illustrations and/or block diagrams of methods, apparatus
(systems) and computer program products according to embodiments of
the invention. It is understood that each block or a portion of the
blocks of the flowchart, illustrations, and/or block diagrams, can
be implemented by computer program instructions in form of computer
executable code when applicable. It is further under stood that,
when not mutually exclusive, combinations of blocks in different
flowcharts, illustrations, and/or block diagrams may be combined.
These computer program instructions may be provided to a processor
of a general purpose computer, special purpose computer, or other
programmable data processing apparatus to produce a machine, such
that the instructions, which execute via the processor of the
computer or other programmable data processing apparatus, create
means for implementing the functions/acts specified in the
flowchart and/or block diagram block or blocks.
These computer program instructions may also be stored in a
computer readable medium that can direct a computer, other
programmable data processing apparatus, or other devices to
function in a particular manner, such that the instructions stored
in the computer readable medium produce an article of manufacture
including instructions which implement the function/act specified
in the flowchart and/or block diagram block or blocks.
The computer program instructions may also be loaded onto a
computer, other programmable data processing apparatus, or other
devices to cause a series of operational steps to be performed on
the computer, other programmable apparatus or other devices to
produce a computer implemented process such that the instructions
which execute on the computer or other programmable apparatus
provide processes for implementing the functions/acts specified in
the flowchart and/or block diagram block or blocks.
A `user interface` as used herein is an interface which allows a
user or operator to interact with a computer or computer system. A
`user interface` may also be referred to as a `human interface
device.` A user interface may provide information or data to the
operator and/or receive information or data from the operator. A
user interface may enable input from an operator to be received by
the computer and may provide output to the user from the computer.
In other words, the user interface may allow an operator to control
or manipulate a computer and the interface may allow the computer
indicate the effects of the operator's control or manipulation. The
display of data or information on a display or a graphical user
interface is an example of providing information to an operator.
The receiving of data through a keyboard, mouse, trackball,
touchpad, pointing stick, graphics tablet, joystick, gamepad,
webcam, headset, pedals, wired glove, remote control, and
accelerometer are all examples of user interface components which
enable the receiving of information or data from an operator.
A `hardware interface` as used herein encompasses an interface
which enables the processor of a computer system to interact with
and/or control an external computing device and/or apparatus. A
hardware interface may allow a processor to send control signals or
instructions to an external computing device and/or apparatus. A
hardware interface may also enable a processor to exchange data
with an external computing device and/or apparatus. Examples of a
hardware interface include, but are not limited to: a universal
serial bus, IEEE 1394 port, parallel port, IEEE 1284 port, serial
port, RS-232 port, IEEE-488 port, Bluetooth connection, Wireless
local area network connection, TCP/IP connection, Ethernet
connection, control voltage interface, MIDI interface, analog input
interface, and digital input interface.
A `display` or `display device` as used herein encompasses an
output device or a user interface adapted for displaying images or
data. A display may output visual, audio, and or tactile data.
Examples of a display include, but are not limited to: a computer
monitor, a television screen, a touch screen, tactile electronic
display, Braille screen, Cathode ray tube (CRT), Storage tube,
Bi-stable display, Electronic paper, Vector display, Flat panel
display, Vacuum fluorescent display (VF), Light-emitting diode
(LED) displays, Electroluminescent display (ELD), Plasma display
panels (PDP), Liquid crystal display (LCD), Organic light-emitting
diode displays (OLED), a projector, and Head-mounted display.
Magnetic Resonance (MR) data is defined herein as being the
recorded measurements of radio frequency signals emitted by atomic
spins using the antenna of a magnetic resonance apparatus during a
magnetic resonance imaging scan. Magnetic resonance data is an
example of medical image data. A Magnetic Resonance Imaging (MRI)
image is defined herein as being the reconstructed two or three
dimensional visualization of anatomic data contained within the
magnetic resonance imaging data. This visualization can be
performed using a computer.
In one aspect the invention provides for a magnetic resonance
imaging system comprising a radio-frequency system for acquiring
magnetic resonance data from an imaging zone. The radio-frequency
system comprises multiple antenna elements. The magnetic resonance
imaging system further comprises a memory containing
machine-executable instructions and pulse sequence commands. The
pulse sequence commands cause the processor to acquire magnetic
resonance data from the multiple antenna elements according to a
SENSE protocol. The SENSE or sensitivity encoding method is a
well-known parallel-imaging reconstruction method used in magnetic
resonance imaging. The SENSE protocol is for example reviewed in
the previously mentioned textbook, Handbook of MRI Pulse Sequences
by Bernstein et al in chapter 13, section 3.
The magnetic resonance imaging system further comprises a
processor. Execution of the machine executable instructions causes
the processor to control the magnetic resonance imaging system with
the pulse sequence commands to acquire the magnetic resonance data.
Execution of the machine-executable instructions further causes the
processor to reconstruct a preliminary image using the magnetic
resonance imaging data. In various examples the preliminary image
could take different forms. For example the pulse sequence commands
may also contain instructions which cause the processor to control
the magnetic resonance imaging system to acquire a survey or scout
image before the SENSE image data is acquired. In other examples
the preliminary image may be a SENSE image reconstructed using a
SENSE imaging protocol.
Execution of the machine-executable instructions further causes the
processor to calculate a fit or registration between an anatomical
model and the preliminary image. The anatomical model may comprise
detailed anatomical data which is registered or fit to the
preliminary image. In other instances the anatomical model may be a
set of procedures or operations which are used to identify regions
of the preliminary image. For example the anatomical model can be
used to identify a region a predetermined distance from an anterior
edge of a subject. The identification of this region may considered
to be a fit or registration.
The anatomical model may be fit or registered to the preliminary
image. Various types of anatomical models include a model which
identifies anatomical landmarks, a model which uses anatomical
atlas compiled from many different subjects, and a deformable shape
model which is deformed to fit the particular preliminary image.
The anatomical model comprises a motion likelihood map. The motion
likelihood map may have data which represents a likelihood that a
particular region identified by the anatomical model could be in
motion during the acquisition of the magnetic resonance data. For
example a region such as a diaphragm, a lung or a heart would be
very likely to be in motion. In other example certain blood vessels
may be marked which may be the origin of potential flow artifacts,
for example due to the fresh inflow of blood. These could be given
a higher value or relative value than other areas in the anatomical
model. In some examples the motion likelihood map may be a
probability distribution, however the motion likelihood map may
also contain a simple weighting which is not normalized or
scaled.
Execution of the machine-executable instructions further causes the
processor to identify at least one image artifact origin at least
partially using the motion likelihood map and the fit or
registration. The fit or registration correlates the position in
the anatomical model to the preliminary image. This enables the
motion likelihood map to be applied to the preliminary image. The
combination of a motion likelihood map and the fit or registration
therefore allow the identification of regions within the
preliminary image which are likely to cause or be the source of
motion artifacts. In some examples the motion likelihood map may be
applied directly to the preliminary image to identify the at least
one image artifact origin. In other examples the motion likelihood
map is used as a weighting in a search algorithm which is used to
identify the at least one image artifact origin.
Execution of the machine-executable instructions further cause the
processor to determine an extended SENSE equation at least
partially using the at least one image artifact origin. During
SENSE image reconstruction images acquired from each of the
multiple antenna elements are reconstructed and then combined,
unfolded into a single image. The extended sense equation may be
represented as a linear system of equations where the sensitivity
matrix has had its rank increased with additional entries to
suppress or eliminate the ghosting or motion artifact. For example
the sensitive matrix could have one or more additional columns
added. Execution of the machine-executable instructions further
causes the processor to construct a corrected SENSE image using the
extended SENSE set of equations. This embodiment may have the
benefit of providing for a means of effectively reducing the
effects of image artifacts in SENSE reconstructed images. It may
have the advantage of providing for reduced motion due to prior
knowledge which is incorporated in the motion likelihood map of the
anatomical model.
The at least one image artifact origin may in some instances be
considered to be a physical location or a location which
corresponds to a physical location in reference to the preliminary
image or a later SENSE image reconstruction. For example if the
preliminary image is a survey or scout image the location in
particular voxels may be different but the spatial location
identified by the anatomical model can be used nonetheless in a
SENSE image reconstruction.
In another embodiment execution of the machine-executable
instructions cause the processor to reconstruct a measured coil
image for each of the multiple antenna elements using the magnetic
resonance data. Execution of the machine-executable instructions
further cause the processor to construct a preliminary SENSE image
by using the set of coil sensitivities to combine the measured coil
image of each of the multiple antenna elements according to the
SENSE protocol. This may be beneficial because the preliminary
SENSE image can be used for identifying the at least one image
artifact origin.
In another embodiment the preliminary SENSE image is performed
using an overdetermined reconstruction. In this case the number of
the multiple antenna elements is greater than the SENSE factor plus
a number of additional constraints. For example if there are 16
antenna elements, the SENSE factor is 3, and for every x-value
there are 2 locations identified as image artifact origins then the
SENSE reconstruction is over determined since 16 is greater than 3
plus 2.
In another embodiment the preliminary image comprises the
preliminary SENSE image. In other words the preliminary image may
be the preliminary SENSE image or the preliminary SENSE image may
be part of image data which is used to make the preliminary
image.
In another embodiment execution of the machine-executable
instructions further cause the processor to construct a
backprojected image for each of the multiple antenna elements using
the preliminary SENSE image and the coil sensitivities. During a
SENSE reconstruction the individual images from each of the antenna
elements are combined using the coil sensitivities. In back
projection the preliminary SENSE image or the reconstructed SENSE
image is used with the coil sensitivities to calculate an image for
each of the multiple antenna elements. If the magnetic resonance
data acquired and the coil sensitivities were perfect then the back
projected image and the measured coil image would be identical.
This is however commonly not the case. There may be errors in the
coil sensitivity matrix that cause differences between the
backprojected image and the measured coil image. These errors may
cause the preliminary SENSE image or portions of the preliminary
SENSE image to become corrupted. Comparing the back projected image
for each of the multiple antenna elements and its measured coil
image therefore provides a way of evaluating how successful the
SENSE reconstruction is.
Execution of the machine-executable instructions further cause the
processor to compare the back projected image to the measured coil
image for each of the multiple antenna elements to identify a set
of affected voxels for each of the multiple antenna elements. The
back projected image may be compared on a voxel to voxel basis to
the measured coil image. If a voxel in the back projected image
varies by more than a predetermined amount or by according to a
statistical measure from measured coil image than that voxel can be
appended to the set of affected voxels. A voxel of set of affected
voxels is a particular voxel that differ in the back projected
image and the measured coil image sufficiently to indicate that
there may be an error in that voxel or in the corresponding voxel
of the preliminary SENSE image.
The identification of the at least one image artifact origin is
performed in image space. The identification of the at least one
image artifact origin is performed at least partially using a set
of affected voxels, at least partially using the motion likelihood
map, and the fit or registration. The set of affected voxels may
for instance be identified by comparing voxel values between the
back projected image and its corresponding measured coil image. If
the value of a particular voxel varies more than a predetermined
threshold then it may be added to the set of affected voxels. This
method may have the benefit that the set of affected voxels, the
motion likelihood map and the fit or registration may be useful for
identifying in space the origin of artifacts caused in the
preliminary SENSE image. For example the set of affected voxels may
be a ghost image or ghosting caused by motion. The combination of
the three may be used to identify the at least one image artifact
origin.
In another embodiment the at least one image artifact origin is
corrected by numerically searching for a maximum of a consistency
measure within a predetermined vicinity of each of the at least one
image artifact origin before constructing the corrected SENSE
image. The use of a consistency measure is detailed in Winkelmann
et al. The consistency measure corresponds to the consistency check
which is illustrated in FIG. 3 of Winkelmann et al and one example
of the consistency measure is equation 7 in Winkelmann et al. The
difficulty in using the ghost artifact removal illustrated in
Winkelmann et al is that it is extremely numerically demanding.
Examples as described herein may have the advantage that the prior
identification of the likely motion areas in the motion likelihood
map may be used to drastically accelerate the numerical
process.
The consistency measure is dependent upon the difference between
the set of affected voxels in the preliminary SENSE image and the
back projected trial SENSE image for each of the multiple antenna
elements. The back projected trial SENSE images are constructed
from a trial SENSE image. The trial SENSE image is constructed
using a trial SENSE equation. An example of a trial SENSE equation
is equation 3 of Winkelman. There is an additional column which has
been added to the coil sensitivity matrix. It is however not known
where the source of the ghosting comes from. For this reason
different trial SENSE equations are constructed and tried to see
which results in the maximum consistency measure. The present
embodiment may have the advantage that prior knowledge of where the
at least one image artifact origin may be located may be used to
greatly accelerate the search for the extended SENSE equation which
provides the best result. The prior knowledge is expressed in terms
of the anatomical model.
In another embodiment the trial SENSE equation that maximizes the
consistency measure is the extended SENSE equation.
In another embodiment execution of the machine-executable
instructions further cause the processor to modify the at least one
image artifact origin by registering the set of affected voxels to
the preliminary image. For example if the set of affected voxels
identifies a ghosting artifact in the image then it may be possible
to simply register the set of affected voxels directly to the
preliminary image. This may aid in identification of the at least
one image artifact origin. In some instances the motion likelihood
map could also be used to provide trial locations for registering
the set of affected voxels to the preliminary image. This may help
in further accelerating the numerical method.
In another embodiment the extended SENSE equation comprises an
extended coil sensitivity matrix. The extended coil sensitivity
matrix is chosen to minimize a contribution from at least a portion
of the at least one image artifact origin.
This may be performed in different ways. In one example the motion
likelihood map alone may be used to identify regions which are
included in the extended SENSE reconstruction. In one concrete
example the preliminary image may be reconstructed and then the
anatomical model is registered to the preliminary image. The motion
likelihood map is then also therefore registered to the preliminary
image. Regions above a particular value or threshold within the
motion likelihood map are then identified as regions or voxels
within the preliminary image which are likely to be causing motion
artifacts. The artifact-contribution of these regions may then be
suppressed by adding the sensitivities of these locations to the
extended coil sensitivity matrix.
In another embodiment the preliminary image comprises a survey scan
image. The use of a survey scan image may be beneficial in some
instances. For example the survey scan image could be acquired
using a body coil or body antenna instead of performing the SENSE
reconstruction. The survey scan image may be less likely to have
image artifacts within it. This may be useful in registering the
anatomical model.
In another embodiment the at least one image artifact is
two-dimensional or three-dimensional. This may be beneficial
because for example the method detailed in Winkelmann et al is so
numerically intensive that it may be difficult to correct for
two-dimensional or three-dimensional regions causing motion
artifacts.
In another aspect the invention provides for a computer program
product comprising machine-executable instructions for execution by
a processor controlling a magnetic resonance imaging system. The
magnetic resonance imaging system comprises a radio-frequency
system for acquiring magnetic resonance data from an imaging zone.
The radio-frequency system comprises multiple antenna elements.
Execution of the machine-executable instructions causes the
processor to control the magnetic resonance imaging system with
pulse sequence commands to acquire the magnetic resonance data. The
pulse sequence commands cause the processor to acquire magnetic
resonance data from the multiple antenna elements according to a
SENSE protocol. Execution of the machine-executable instructions
further causes the processor to reconstruct a preliminary image
using the magnetic resonance imaging data. Execution of the
machine-executable instructions further causes the processor to
calculate a fit or registration between an anatomical model and the
preliminary image.
The anatomical model comprises a motion likelihood map. Execution
of the machine-executable instructions further causes the processor
to identify at least one image artifact origin at least partially
using the motion likelihood map and the fit or registration.
Execution of the machine-executable instructions further causes the
processor to determine an extended SENSE equation at least
partially using the at least one image artifact origin. Execution
of the machine-executable instructions further cause the processor
to construct a corrected SENSE image using the extended SENSE
equation. The advantages of this have been previously
discussed.
In another aspect the invention provides for a method of operating
a magnetic resonance imaging system. The magnetic resonance imaging
system comprises a radio-frequency system for acquiring magnetic
resonance data from an imaging zone. The radio-frequency system
comprises multiple antenna elements. The method comprises the step
of controlling the magnetic resonance imaging system with the pulse
sequence commands to acquire the magnetic resonance data. The pulse
sequence commands cause the processor to acquire magnetic resonance
data from the multiple antenna elements according to a SENSE
protocol. The method further comprises the step of reconstructing a
preliminary image using the magnetic resonance imaging data.
The method further comprises the step of calculating a fit or
registration between an anatomical model and the preliminary image.
The anatomical model comprises a motion likelihood map. The method
further comprises the step of identifying at least one image
artifact origin at least partially using the motion likelihood map
and the fit or registration. The method further comprises the step
of determining an extended SENSE equation at least partially using
the at least one image artifact origin. The method further
comprises the step of constructing a corrected SENSE image at least
partially using the extended SENSE equation.
It is understood that one or more of the aforementioned embodiments
of the invention may be combined as long as the combined
embodiments are not mutually exclusive.
BRIEF DESCRIPTION OF THE DRAWINGS
In the following preferred embodiments of the invention will be
described, by way of example only, and with reference to the
drawings in which:
FIG. 1 illustrates an example of a magnetic resonance imaging
system;
FIG. 2 shows a flow chart which illustrates a method of operating
the magnetic resonance imaging system of FIG. 1;
FIG. 3 illustrates a further example of a magnetic resonance
imaging system;
FIG. 4 shows a flow chart which illustrates a method of operating
the magnetic resonance imaging system of FIG. 3;
FIG. 5 illustrates the Scheme of an annotated body model guided
extended SENSE reconstruction;
FIG. 6 diagrammatically illustrates the effect of motion in a
magnetic resonance image;
FIG. 7 further diagrammatically illustrates the effect of motion in
a magnetic resonance image;
FIG. 8 diagrammatically illustrates the choosing of a region which
is minimized during an extended SENSE reconstruction to reduce
motion artifacts;
FIG. 9 shows a magnetic resonance image that has been modified to
simulate motion in a fat layer; and
FIG. 10 shows a simulation of removing motion artifacts from the
image shown in FIG. 9.
DETAILED DESCRIPTION OF THE EMBODIMENTS
Like numbered elements in these figures are either equivalent
elements or perform the same function. Elements which have been
discussed previously will not necessarily be discussed in later
figures if the function is equivalent.
FIG. 1 shows an example of a magnetic resonance imaging system 100
with a magnet 104. The magnet 104 is a superconducting cylindrical
type magnet 104 with a bore 106 through it. The use of different
types of magnets is also possible for instance it is also possible
to use both a split cylindrical magnet and a so called open magnet.
A split cylindrical magnet is similar to a standard cylindrical
magnet, except that the cryostat has been split into two sections
to allow access to the iso-plane of the magnet, such magnets may
for instance be used in conjunction with charged particle beam
therapy. An open magnet has two magnet sections, one above the
other with a space in-between that is large enough to receive a
subject: the arrangement of the two sections area similar to that
of a Helmholtz coil. Open magnets are popular, because the subject
is less confined. Inside the cryostat of the cylindrical magnet
there is a collection of superconducting coils. Within the bore 106
of the cylindrical magnet 104 there is an imaging zone 108 where
the magnetic field is strong and uniform enough to perform magnetic
resonance imaging.
Within the bore 106 of the magnet there is also a set of magnetic
field gradient coils 110 which is used for acquisition of magnetic
resonance data to spatially encode magnetic spins within the
imaging zone 108 of the magnet 104. The magnetic field gradient
coils 110 are connected to a magnetic field gradient coil power
supply 112. The magnetic field gradient coils 110 are intended to
be representative. Typically magnetic field gradient coils 110
contain three separate sets of coils for spatially encoding in
three orthogonal spatial directions. A magnetic field gradient
power supply supplies current to the magnetic field gradient coils.
The current supplied to the magnetic field gradient coils 110 is
controlled as a function of time and may be ramped or pulsed.
Within the bore 106 of the magnet 104 is an optional body coil 114.
The body coil 114 may also be referred to as a body antenna. The
body coil 114 is shown as being connected to a transceiver 116. In
some embodiments body coil 114 may also be connected to a whole
body coil radio frequency amplifier and/or receiver, however this
is not shown in this example. If both a transmitter and a receiver
are connected to the whole body coil 114, a means for switching
between the transmit and receive mode may be provided. For example
a circuit with a pin diode may be used to select the transmit or
receive mode. A subject support 120 supports a subject 118 within
the imaging zone.
A transceiver 122 is shown as being connected to a magnetic
resonance imaging antenna 124. In this example the magnetic
resonance imaging coil 124 is a surface coil comprising multiple
antenna elements 126, 126', 126'', 126'''. The transceiver 122 is
operable for sending and receiving individual RF signals to the
individual coil elements 126, 126', 126'', 126'''. In this example
the transceiver 116 and the transceiver 122 are shown as being
separate units. However, in other examples the units 116 and 122
could be combined.
The magnetic resonance imaging system comprises a computer system
130. The transceiver 116, the transceiver 122, and the magnetic
field gradient coil power supply are shown as being connected to a
hardware interface 132 of the computer 130. The computer 130 is
further shown as containing a processor 134 which is operable for
executing the machine-readable instructions. The computer 130 is
further shown as comprising a user interface 136, computer storage
138 and computer memory 140 which are all accessible and connected
to the processor 134.
The computer memory 138 is shown as containing pulse sequence
commands 150. The pulse sequence commands contain instructions
which enable the processor 134 to control the magnetic resonance
imaging system 100 to acquire magnetic resonance data 152 according
to a SENSE protocol. The pulse sequence commands may contain
instructions which enable the processor 134 to acquire magnetic
resonance data according to more than one imaging protocol. The
pulse sequence commands 150 enable the processor 134 to acquire
data according to a SENSE protocol but it may also enable other
protocols to be used such as acquiring a survey or scout scan
before performing the SENSE protocol. The computer storage 138 is
shown as containing magnetic resonance data 152 which was acquired
using the pulse sequence commands 150 to control the
acquisition.
The computer storage 138 is further shown as containing a
preliminary image 154 that was reconstructed from the magnetic
resonance data 152. The computer storage 138 is further shown as
containing an anatomical model 156 which can be fit or registered
to the preliminary image 154. The computer storage 138 is further
shown as containing a motion likelihood map 158 that is registered
or is part of the anatomical model 156. The computer storage 138 is
further shown as containing a fit 159 or registration that was
calculated between the preliminary image 154 and the anatomical
model 156. The computer storage 138 is further shown as containing
a location of an image artifact origin 160 that was identified in
the preliminary image 154. The location of the image artifact
origin 160 is referenced to a location or set of locations in the
preliminary image 154. If the magnetic resonance imaging system 100
is properly calibrated then the location of the image artifact
origin 160 may also be stored in terms of coordinates relative to
the subject 118.
The computer storage 138 is further shown as containing an extended
SENSE equation 162. The extended SENSE equation contains one or
more additional columns in its sensitivity matrix that are used to
minimize sensitivity of the reconstruction-result to motion at the
location of the image artifact origin. The computer storage 138 is
further shown as containing a corrected SENSE image 164. The
computer memory 140 is shown as having machine-executable
instructions 170. The machine-executable instructions 170 contain
instructions which enable the processor 134 to control the
operation of the magnetic resonance imaging system 100 and also to
perform image reconstruction and modifications of magnetic
resonance data and magnetic resonance images. The
machine-executable instructions 170 may cause the processor 134 to
perform a computer implemented method as is described in any one of
the following method flowcharts.
FIG. 2 shows a flowchart which illustrates a method of operating
the magnetic resonance imaging system 100 of FIG. 1. The following
method steps could be implemented by the machine-executable
instructions 170 shown in FIG. 1.
First in step 200 the magnetic resonance imaging system is
controlled with the pulse sequence commands 150 to acquire the
magnetic resonance data 152. Next in step 202 a preliminary image
154 is reconstructed from the magnetic resonance data 152. Next in
step 204 a fit or registration between an anatomical model 156 and
the preliminary image 154 is calculated. The anatomical model 156
comprises a motion likelihood map 158. In step 206 at least one
image artifact origin 160 is identified at least partially using
the motion likelihood map 158 and the fit or registration 159. Next
in step 208 an extended equation 162 is constructed at least
partially using knowledge of the at least one image artifact origin
160. Finally in step 210 a corrected SENSE image 164 is constructed
according to the extended SENSE equation 162.
In one variation, the extended SENSE equation is constructed by
adding one or more additional columns to the coils sensitivity
matrix to construct an extended coil sensitivity matrix. The
extended coil sensitivity matrix is chosen to minimize a
contribution from at least a portion of the at least one image
artifact origin.
FIG. 3 shows a further example of a magnetic resonance imaging
system 300. The system 300 is similar to the system 100 shown in
FIG. 1. The computer storage 138 is further shown as containing
measured coil images 302 that were reconstructed from the magnetic
resonance data 152. The computer storage 138 further shows a
preliminary SENSE image 304 that was constructed using the measured
coil images 302 and the set of coil sensitivities 306 which are
shown as being stored in the computer memory 138. The set of coil
sensitivities 306 could be known a priori or they may be measured
during a SENSE calibration step. For example a survey image could
be acquired using the body coil and be used to calibrate each of
the antenna elements 126, 126', 126'', 126'''.
The computer storage 138 is further shown as containing back
projected images 308 that were calculated from the preliminary
SENSE image 304 and the set of coil sensitivities 306. The back
projected images 308 are used to create images that show what the
measured coil images 302 would be like if the set of coil
sensitivities 306 were perfectly known and also there were no image
artifacts. The back projected images 308 are however different from
the measured coil images 302. Comparing the measured coil image and
the back projected image for each of the coil elements 126, 126',
126'', 126''' a set of affected voxels 310 can be identified for
each coil. These may be used to identify image artifacts within the
measured coil images. These also may lead to the identification of
image artifacts within the preliminary SENSE image 304. The
computer memory 140 is again showing the machine-executable
instructions 170. The machine-executable instructions could for
instance cause the processor 134 to perform a computer-implemented
method as described in FIG. 2 or also in the following FIG. 4.
FIG. 4 shows a flow diagram which illustrates a method similar to
that illustrated in FIG. 2. In FIG. 4 a number of additional steps
have been added to the method. FIG. 4 starts with steps 200 and 202
as are described in FIG. 2. Next in step 400 a measured coil image
is reconstructed for each of the multiple antenna elements 126,
126', 126'', 126''' using the magnetic resonance data 152. Next in
step 402 a preliminary SENSE image 304 is constructed by using the
set of coil sensitivities 306 to combine the measured coil image
302 for each of the multiple antenna elements 126, 126', 126'',
126''' according to a SENSE magnetic resonance imaging protocol. In
some examples the preliminary SENSE image 304 may be the
preliminary image 154. In this case step 204 is now performed
because step 204 is identical with steps 400 and 402.
In other examples however the preliminary image is different or
distinct from the preliminary SENSE image 304. For example the
pulse sequence commands 150 could cause the magnetic resonance
imaging system 100 to acquire a survey or scout scan and this may
then be fit to the anatomical model. In yet other examples the
survey or scout scan image and also the preliminary SENSE image are
both used for fitting to an anatomical model while performing the
computer implemented method.
As described above the method then may optionally perform step 204.
In the method shown in FIG. 4 steps 404, 406, 408 are more detailed
instructions on how step 206 of FIG. 2 is performed. First in step
404 a back projected image for each of the multiple antenna
elements 126, 126', 126'', 126''' is constructed using the
preliminary SENSE image 304 and the set of coil sensitivities 306.
Next in step 406 the back projected image 308 is compared to the
measured coil image 302 for each of the multiple antenna elements
126, 126', 126'', 126''' to identify a set of affected voxels 310.
The identification of the at least one image artifact origin is
then performed at least partially using the set of affected voxels
310 and the motion likelihood map 158 and the fit or registration
159. Next in step 308 the at least one image artifact origin 160 is
corrected by numerically searching for a maximum of the consistency
measure within a predetermined vicinity of each of the at least one
image artifact origin before constructing the corrected SENSE image
164.
For example the numerical method described in Winkelmann et. al.
could be applied where the search area is limited to the
predetermined vicinity for a particular voxel. In this case the
algorithm of Winkelmann et all is applied but the search area has
been greatly reduced using processing in image space. This may lead
to a dramatic increase in the numerical efficiency in finding the
at least one image artifact origin. After the steps 404, 406, and
408 may be repeated numerically during the process of searching for
a maximum of a consistency measure. Next steps 208 and 210 are
performed as is described in FIG. 2.
SENSE is one of the techniques to accelerate Magnetic Resonance
(MR) data acquisition. To facilitate the unfolding, coil
sensitivity information and measured data have to be available as
the essential inputs for the SENSE algorithm. If coil sensitivities
are perfect, SENSE image artifacts might still appear due to errors
in the measured data. The most prominent artifact of this kind,
caused by a distortion of the Fourier encoding process, is
ghosting, resulting in a displaced appearance of parts of the
actual signal. However, although the Fourier encoding process is
spoiled the actual signal is properly sensitivity encoded which
gives a handle to identify and to remove the ghosting structures,
see the Winkelmann et. al. paper.
Key of this idea is to identify signal at risk by a consistency
check and to solve an extended SENSE problem including a displaced
signal component (the ghost) for which the right origin has to be
found in a search procedure performed along the phase encoding
direction. The numerical burden of this search might be feasible in
1D under-sampling but gets problematic in two dimensions and could
run into false minima in some poor cases. In some examples, the
idea is to incorporate an appropriate, patient-adapted body model
into the extended SENSE reconstruction to accelerate and stabilize
the search for the ghost origin by incorporating specific prior
knowledge. Based on 3D scout MRI data (or other images), acquired
at the beginning of the examination, an appropriate body model (or
anatomical model) can be adapted to the patient. The model may
contain annotated organs (i.e., liver, lung, etc.) and structures
like vessels, fluid-filled chambers (heart, etc.), chest-wall and
so on which have a high probability to be the origin of MR ghosts
(e.g. due to inflow, flow, motion effects, etc. . . . ). This model
adapted to the folded SENSE/consistency data can guide the ghost
origin search procedure decreasing the numerical effort and
increasing confidence.
SENSE is the method for parallel imaging. If coil sensitivity
information is perfect, SENSE image quality depends still on the
measured data. Those can contain some data inconsistencies e.g.
caused by motion like bulk motion, flow and fresh magnetization
in-flow, etc. which actually can spoil the Fourier encoding
process. Ghosting, apart from blurring, is the most prominent image
artifact, which can't be resolved by a standard SENSE
reconstruction.
However, it was found that the ghost, although shifted in the phase
encoding direction (due to a corrupted Fourier encoding process) in
the final SENSE image, is properly sensitivity encoded. However,
its actual location, the spot where it came from, is unknown (1).
Voxels corrupted by ghosting signals can therefore be found in the
reconstructed SENSE image by analyzing the consistency between the
reconstructed SENSE image and the underlying folded data (1). The
consistency test shows the ghost because it was exposed to a
different receive coil sensitivity during signal detection compared
to what has been assumed during SENSE reconstruction.
To remove the ghost for those corrupted voxels an extended SENSE
problem must be formulated asking for an additional signal
contribution; the ghost--stemming from another location, folding
onto the corrupted voxel. The resulting signal model differs from
that of normal SENSE by the additional term, S.sub.i,g .delta., in
Eq. [1] of Winkelmann et. al.
.SIGMA..times..times..rho..times..delta. ##EQU00001##
Here the vector C contains the measured folded coil signals for the
coil i, the measurement data. S is denoting the sensitivity matrix,
.rho. is the vector containing the actual voxel signals to be
obtained, .delta. is the ghost signal contribution coming from the
unknown location g while the sum run over all coils. This extended
SENSE problem can only be established if the SENSE problem is over
determined. Having one phase encoding direction (2D imaging)
increases the rank of the extended SENSE matrix by one (only one
ghost is expected) having two directions like in 3D imaging could
result in two ghosting sources, increasing the rank of the extended
SENSE matrix even further, with the consequence to reduce the
condition number of the pseudo inverse making solution prone to
noise. The extended SENSE problem is key element of an optimization
to obtain best data consistency (firming as a penalty term) with
respect the potential location g of the ghost. This search might be
feasible in 1D under-sampling (see FIG. 3a of Winkelmann et. al.)
but gets problematic in two dimensions and could run into false
minima in some poor cases.
In FIG. 3a of Winkelmann et. al. a one-dimensional search along the
phase encoding direction for a ghosting artifact is 2D imaging. The
consistency log(P) of the extended SENSE reconstruction plotted as
a function of the potential origin of the ghost .delta. along the
phase encoding direction. Two different ghosting voxels in the
final SENSE image are shown (I, II). At the ghost appearance
(LAA--location of artifact appearance in the final image) the
problem is singular at the ghost origin (LAO--location of artifact
origin) the consistency shows a maximum. For ghost (I) the main
maximum is roughly well defined, for ghost (II) the problem might
be harder.
Furthermore, the condition of the inverse of Eq. [1] gets poor if
the origin of the artifact g is close to the location I of the
voxel under study making the extended SENSE solution more
difficult.
To speed up and to stabilize the solving process of the extended
SENSE problem Eq. [1] it is proposed to incorporate prior
knowledge. This is realized by using an appropriate,
patient-adapted body model in the extended SENSE reconstruction.
This model indicates potential regions at risk to be a source of
ghosting artifacts helping to guide the search and to rule out
potential false positives.
Based on the 3D scout MRI data, acquired at the beginning of the
examination, the body model can be adapted patient specifically,
reflecting major features of the patient's anatomy and geometry
appropriately. The model contains annotated organs (liver, lung,
etc.) and structures (like vessels, fluid-filled chambers (the
heart, etc.), chest-wall, etc.) which have a high probability to be
the origin of MR ghosts (e.g. due to inflow, flow, motion effects,
etc.).
Models can be adapted to the extended SENSE/consistency data, can
guide the ghost origin search procedure decreasing the numerical
effort and increase confidence.
Furthermore, the adapted body model might be helpful to guide the
SENSE signal model also in other directions. Based on the adapted
body model, areas in the coil sensitivity maps can be identified
which are at risk to be wrong. These areas can also be added as
columns to the sensitivity matrix, further extending the
sensitivity matrix.
In one example, a liver examination is performed on a subject or
patient. The liver is surrounded by a corresponding multi-element
reception coil, is placed in the iso-center of the magnet. A
multi-slice or 3D low resolution scout scan or other scan is
measured for planning purposes. The data are further send to an
image-processing algorithm that fits a predefined and body model of
the corresponding body region to the data using elastic
registration. The identification of the coarse body region could be
done automatically or could be context driven. After the
registration process the patient geometry is matched to the
information about the risk for motion artifacts and their nature.
This information will become available for each voxel within the
patient and will be accessible for all algorithms to be performed
subsequently.
Coil sensitivity information is obtained using a SENSE reference
scan. Using the receiver array a 3D breath-hold (15-20 sec.)
diagnostic liver scan is performed covering the entire abdominal
region. Due to cardiac motion fresh blood is pumped into the 3D
volume, scanned, resulting due in-flow effects into an intensity
modulation of the MR signal causing a Fourier ghosting artifact in
the sub-sampled SENSE imaging. This artifact also propagates during
SENSE reconstruction into the entire FOV.
This artifact is identified by the consistency check proposed in
Winkelmann et. al. This check measures how the measured,
sub-sampled reduced FOV images fit the finally SENSE reconstructed
images.
To facilitate this comparison the final SENSE image is
back-projected onto the individual reduced FOV coil images for each
channel; their difference to the measured data has to match the
receiver noise level and is evaluated using an appropriate
probability (see Winkelmann et. al.). Not matching allows
identification of corrupted voxels that need a different solution
to the SENSE problem employing the extended SENSE approach (see
Winkelmann et. al.). Here the geometrically matched model comes
into play, briefly illustrated in FIG. 5 below, which
supports/guides the search along the corresponding phase encoding
direction, performed to find the origin of the ghost. Thus, the
extended SENSE approach is augmented by the annotated body model
which informs the algorithm about which pixels in the full Field of
View (FOV) or in the respective sub-sampled FOV are potential
sources of artifacts. Those and their neighborhood will be tested
regarding the achievable consistency assuming that they are the
source. Furthermore, the information from the model can be used the
additionally regularize potential others solutions (locations) to
avoid false solutions.
This approach speeds up the algorithm and can be used to avoid
noise break though or false positives.
FIG. 5 shows an idealized representation of a preliminary SENSE
image 304, an anatomical model 156, and a corrected SENSE image
164. The preliminary SENSE image 304 shows a cross-section of a
subject and there are a number of ghost artifacts 500 of an aorta
502. The movement of blood within the aorta 502 causes the ghost
artifacts 500. The anatomical model 156 may have a motion
likelihood map 158 assigned to it. In this map the aorta 502 is
identified as a region which has a high likelihood of causing a
ghosting artifact. This model 156 is then fit to the preliminary
SENSE image 304 and the method as described in FIGS. 3 and 4 is
applied to calculate the corrected SENSE image 164. FIG. 5
illustrates the Scheme of an annotated body model guided extended
SENSE reconstruction. The SENSE reconstructed MR image shows
ghosting artifacts (replica of the pulsating aorta in phase
encoding direction, vertical) which are identified by the
consistency check. The geometrically matched, adapted body model
(middle) shows the potential origin of this ghosting artifact
(aorta--high risk area highlighted in red). This information is
used to guide the search of the extended SENSE reconstruction
removing the artifacts schematically shown (right).
Such a model can furthermore serve many other purposes, for
example: Organ specific planning Automatic navigator positioning
Automatic shim volume positioning Automatic outer-volume
suppression (REST-slab) positioning.
In MRI, it can frequently happen that a region that is not of
clinical interest disseminates artifacts over clinically
interesting regions. One typical example is the aorta that
disseminates flow-artifacts onto the liver; another one is the
breast or the heart, which, by motion, may cause artifacts over the
spine. For an example, see FIG. 6 below. It schematically shows a
sagittal view through the human body, containing a spine that is of
clinical interest and a heart that is not of interest in that
study.
FIG. 6 shows a sagittal view 600 showing the heart 602. As the
heart is moving it causes image artifacts 604 or ghosting images of
the heart 602.
In some cases, the problem can be overcome by positioning so-called
REST (for "Regional Saturation Technique") over the problematic
(usually moving) body area. However, this often is bound to
restrictions, e.g., regions other than straight slabs are usually
impractical. It also can have serious drawbacks in terms of scan
time, attainable repetition time etc.
With a large numbers of receive elements, this problem can be
solved in a completely different way. The coil combination
algorithm can be made such that the result, in combination, is
minimally sensitive to an area that is known to cause trouble.
This could be done by the user indicating an area known for causing
motion-artifacts. This has some similarity to the placement of
REST-slabs, which is also planned by the user. As a difference, a
motion-presence indication for the sake of optimal
synergy-combination can be made before or after the actual
measurement (assuming the raw data of the measurement is still kept
in memory). For that reason, the technique is called "post-scan
REST".
However, the necessity for user-input hampers workflow. It is
therefore an aim of the present invention to apply the same
principle, but interface-free.
Particularly in abdominal imaging, we assume that motion-artifacts
are predominantly caused by the anterior subcutaneous fat of the
patient. The idea is to detect that region and to devise, in the
reconstruction process, a coil-combination algorithm that minimizes
sensitivity to that region.
FIG. 7 shows a Fig. which represents an abdominal section MRI. The
region labeled 702 represents fat. The region labeled 704
represents an organ of interest such as the liver. The various
lines 706 represent image artifact or ghosting artifacts caused by
movement of the fat layer 702.
In some examples the method applies to sagittal or axial abdominal
imaging that is done with a multiplicity of receive antennas or
antenna elements. In principle, the invention is compatible with
all types of acquisition sequences and does not require
modification thereof. The invention consists of a modified
reconstruction `regular` data.
One element of some examples is the detection of the anterior edge
of the anterior subcutaneous-fat region. This can be done by
analyzing an initial reconstruction of the scan at hand, an image
of the same region from a previous scan or a coil-reference scan or
the like.
Optionally, an estimate of the thickness of the fat-layer can be
provided by a low-resolution chemical shift imaging scan (such as
water-fat separating scan; silicon-fat separating scan);
alternatively, a `typical` fat-layer thickness can be
pre-programmed. Fully automated image segmentation, such as
intensity-based or atlas-based segmentation, may also serve to
generate the initial estimate of the object whose signal intensity
is to be suppressed.
Given an initial estimate of the object, the most anterior
object-part is considered as the `problematic` fat-layer. The
anterior edge combined with the thickness provides us with the
region (a 2D region in a slice, or a 3D region in a multiplicity of
slices.
Alternatively, we can determine the position of a curve (in
multi-slice or 3D: a plane) representing the center of the
anterior-fat region.
Given that region, or the center plane, we can devise a
reconstruction that minimizes sensitivity to that region. This is
done by choosing, for each pixel to be reconstructed, the most
appropriate coil-element weighting that is insensitive to the
region of the anterior fat area.
FIG. 8 is similar to that shown in FIG. 7. In this schematic of an
MRI the abdominal section 700 is shown again. In this case there is
a region 800 which the combination of coil element weights is made
insensitive. This eliminates the image artifacts 706 or ghosting
artifacts shown in FIG. 7.
In some examples, reconstruction algorithm comprises appending, to
the coil sensitivity matrix S an extra row s.sub.a (or a
multiplicity of rows s.sub.a1, . . . s.sub.ak, . . . s.sub.aK,
where k runs over all identified locations where motion-artefacts
may originate). This leads to an "extended coil sensitivity matrix"
S.sub.E=[S s.sub.a] (or, alternatively, S.sub.E=[S s.sub.a1 . . .
s.sub.ak . . . s.sub.aK]). With this matrix, a regular solution to
the SENSE-problem can be applied, i.e.:
.times..PSI..times..times..times..PSI..times. ##EQU00002##
Except for the subscript `E`, this looks like a very familiar
SENSE-equation, with p representing a resulting set of pixels,
.PSI. the noise covariance between acquisition channels, R the
regularization matrix and m the measured coil-array data. Yet, both
p and R now contain one (or a series of) extra elements. In this
invention, the SENSE equation is not producing an equidistant set
of unfolded pixels, but the `regular` equidistant set plus an
estimate of the pixel-intensity on the location of the
artefact-source (i.e. the anterior fat rim); within the scope of
this invention, that extra result is regarded as uninteresting.
Similarly, a difference between a `regular` regularization matrix R
and the "R.sub.E", is that the latter also has to indicate the
expected signal level at the location of the anterior fat--which is
equally well known as the other diagonal elements of the matrix
R.
FIGS. 9 and 10 show magnetic resonance images 900 and 1000. FIG. 9
shows a magnetic resonance image 900 where simulated motion
artifacts have been added to the image.
FIG. 10 shows a magnetic resonance image 1000 where a region of
coil weight elements was made insensitive to remove the artificial
motion artifacts of the image 900.
In the following description, "Possup" is used as shorthand for
"Post-scan sensitivity suppression to anterior fat layer". In
essence, the idea is to slightly alter the SENSE reconstruction in
such a way that the resulting antenna element combination is
minimally sensitive to an (automatically detected) region of the
patient, e.g., the anterior subcutaneous fat region in abdominal
imaging.
A breathhold-image of the abdomen was taken as a starting point.
That image showed relatively little "natural" motion artifacts.
Subsequently, four coil sensitivity maps were conceived, decaying
as 1/x from the anterior edge, the right edge, the posterior edge
and the right edge of the image.
Motion was simulated by creating 6 different `distorted` images.
Each distortion left the posterior half of the object unaffected,
but stretched the anterior half such that the upper edge of the
object was moved anteriorly by 1, 2, 3, 4, 5 or 6 pixels. (The
object is approximately 200 pixels in size, so the anterior half is
approximately 100 pixels, which means that the biggest distortion
stretched the anterior half by approximately 6%). The distorted
image is shown in FIG. 9.
All these sets were Fourier-transformed to k-space, and each
ky-line of the simulation was randomly picked from one of these six
sets.
Reconstructing Aforementioned Simulation with CLEAR Results in the
Image to the Right--
Using this image as a first iteration, the anterior edge of the
object was detected. Subsequently, two regions parallel to that
anterior edge were defined: one at a depth of 5 pixels (anteriorly)
and one at a depth of 12 mm. The knowledge of the thickness of the
fat layer was manually entered. This delivers two curves F1(x) and
F2(x).
The Possup reconstruction calculates, for each location (x,y), a
coil-element weighting that is minimally sensitive to regions F1(x)
and F2(x), while maximizing sensitivity to (x,y). It subsequently
combines coil element data.
Not unexpectedly, the anterior-fat region gets suppressed
in-between 5 and 12 pixels of depth. Mathematically, this consists
of extending the SENSE coil-sensitivity matrix with two extra
columns. This is relevant here, because it allows an easy
combination with SENSE. By applying Possup, also this artifact is
largely removed as is shown in FIG. 10.
While the invention has been illustrated and described in detail in
the drawings and foregoing description, such illustration and
description are to be considered illustrative or exemplary and not
restrictive; the invention is not limited to the disclosed
embodiments.
Other variations to the disclosed embodiments can be understood and
effected by those skilled in the art in practicing the claimed
invention, from a study of the drawings, the disclosure, and the
appended claims. In the claims, the word "comprising" does not
exclude other elements or steps, and the indefinite article "a" or
"an" does not exclude a plurality. A single processor or other unit
may fulfill the functions of several items recited in the claims.
The mere fact that certain measures are recited in mutually
different dependent claims does not indicate that a combination of
these measured cannot be used to advantage. A computer program may
be stored/distributed on a suitable medium, such as an optical
storage medium or a solid-state medium supplied together with or as
part of other hardware, but may also be distributed in other forms,
such as via the Internet or other wired or wireless
telecommunication systems. Any reference signs in the claims should
not be construed as limiting the scope.
LIST OF REFERENCE NUMERALS
100 magnetic resonance imaging system 104 magnet 106 bore of magnet
108 imaging zone 110 magnetic field gradient coils 112 magnetic
field gradient coil power supply 114 body coil or body antenna 116
transceiver 118 subject 120 subject support 122 transceiver 124
magnetic resonance image antenna 126 antenna element 126' antenna
element 126'' antenna element 126''' antenna element 130 computer
132 hardware interface 134 processor 136 user interface 138
computer storage 140 computer memory 150 pulse sequence commands
152 magnetic resonance data 154 preliminary image 156 anatomical
model 158 motion likelihood map 159 fit or registration 160
location of image artifact origin 162 extended SENSE equation 164
corrected SENSE image 170 machine executable instructions 200
controlling the magnetic resonance imaging system with pulse
sequence commands to acquire the magnetic resonance data 202
reconstructing a preliminary image using the magnetic resonance
imaging data 204 calculating a fit or registration between an
anatomical model and the preliminary image, wherein the anatomical
model comprises a motion likelihood map 206 identifying at least
one image artifact origin at least partially using the motion
likelihood map and the fit or registration 208 determining an
extended SENSE equation at least partially using at least one image
artifact origin 210 constructing a corrected SENSE image according
to an extended SENSE reconstruction at least partially using the
extended SENSE equation 300 magnetic resonance imaging system 302
measured coil images 304 preliminary SENSE image 306 set of coil
sensitivities 308 back projected images 310 set of affected voxels
400 reconstructing a measured coil image for each of the multiple
antenna elements using the magnetic resonance data 402 constructing
a preliminary SENSE image by using a set of coil sensitivities to
combine the measured coil image for each of the multiple antenna
elements according to the SENSE protocol 404 constructing a
backprojected image for each of the multiple antenna elements using
the preliminary SENSE image and the coil sensitivities 406
comparing the backprojected image to the measured coil image for
each of the multiple antenna elements to identify a set affected
voxels for each of the multiple antenna elements 408 numerically
searching for a maximum of a consistency measure within a
predetermined vicinity of each of the at least one image artifact
origin before constructing the corrected SENSE image to correct the
at least one image artifact origin 500 ghost artifact 502 aorta 600
sagittal view 602 heart 604 image artifacts 700 abdominal section
702 fat 704 organ of interest 706 image artifacts 800 region 900
magnetic resonance image 1000 magnetic resonance image
* * * * *